Table 2

Characteristics of included studies

Study (country)	Aim	Study design summary	Participant type	Types of phenomena of interest	Sampling frame, and recruitment	Method of data collection	Method of data analyses
Rivara et al (USA)24	To explore parental knowledge, attitudes, beliefs and barriers to use of booster seats in cars for 4–8-year-old children.	Focus group discussion (FGD) (three).	Parents of Children aged≤10 years (n=30).	Knowledge, attitudes, beliefs and barriers.	Recruited through fliers at local child and day care facilities. Participants were paid $30 cash and given a $15 gift certificate.	Not adequately reported. FGDs were 2 hour duration.	The results of the FGDs were examined using the PRECEDE-PROCEED model of Green and Kreuter.40
Simpson et al (USA)34	To identify barriers to booster seat use and to identify potential strategies to increase the use of booster seats among young preschool-aged and school-aged children.	FGD and in-depth interviews (IDI) in two phases.	Parents and children (n=111) Participant groupings were determined according to the age of the child, method of restraint. Those who had participated in any FGD in the last 3 months or were employees of the firm which conducted this were excluded.	Perceptions and experiences.	Phase I (n=54): 31 parents in 3 FGD+8 children in 2 FGD+15 parents in IDIs. Phase II (n=57): 36 parents in 4 FGD+21 parents in telephone discussions. Recruited by local market research firms randomly chosen from databases. Eligible candidates received an honorarium for participation.	Moderator introduced the FGD process and the specific topic for the group. Sessions were audiotaped and videotaped and observed through a 1-way mirror by 2 physicians, a psychologist and a qualitative research specialist.	Experts, the moderator and a research assistant conducted data analysis. Transcripts were reviewed for common themes and new ideas. After each phase of FGDs and IDIs, the moderator drafted an initial summary for review. All researchers who observed the FGDs reviewed the summary and made comments to reach consensus. All possible interpretations were recorded when no consensus was reached.
Lee et al (USA)25	To investigate Latino parents’ knowledge, attitudes and beliefs about booster seats, barriers to booster seat use and effective strategies for message delivery in the Latino community.	FGDs (two).	Parents were eligible if they were Spanish-speaking, had children under the age of 10 years and owned or drove a car.	Knowledge attitudes and beliefs.	Recruited by a member of the research team, who went to two community centres in areas with high Latino population in Seattle. There were 23 participants (12 in the first group and 11 in the second).	Not much is reported—article only mentions what topics were discussed.	Only mentions that the results of the FGDs were audiotaped and reviewed by the authors to identify important themes.
Agran et al (USA)26	To examine the factors contributing to non-use or part-time use of a child restraints and the effects of exposure to a violator class for those cited for violation of the California Child Passenger Safety (CPS) law.	FGDs and results were used to guide the development of a study of an existing violator class for CPS law.	3 FGDs included 24 individuals who had been cited for violation of the California CPS law.	Not specified.	Recruited from violators registering for a class required by several courts in Los Angeles County.	2 FGDs in English and 1 in Spanish. Each of 90 min duration just before the violator class. Facilitators familiar with child occupant protection and child development led the groups.	An observer, supplemented by audiotapes of the sessions, conducted a thematic analysis of notes. The University of California institutional review board approved the protocol.
Lennon (Australia)27	To explore parental perceptions of barriers to placing their children in the rear seat of passenger vehicles whenever possible.	Interpretive phenomenological qualitative approach with FGDs being held with urban parent-driver.	Parents of children aged>12 years who regularly drive their children in passenger vehicles (with a rear seat). 24 parents participated in 5 separate groups.	Perception.	Parents of children aged>12 years who regularly drive their children in passenger vehicles (with a rear seat) were recruited by personal approach in the open-air carparks of two urban shopping centres in Brisbane, Australia.	A semistructured discussion schedule was used. Except for one group where notes were taken, discussions were transcribed by a professional stenographer to allow the accurate recall of exact terms used by parents.	Transcripts and discussion notes were analysed using QSR NVivo V.2.0. Thematic analysis, following the process described by Van Manen41 until data saturation was achieved.
Medoff-Cooper and Tulman (USA)28	To examine the beliefs, behaviours, barriers and motivating factors for use of child car seats among children aged 3–7 years.	FGDs and IDIs.	Mothers (16 in 2 FGDs) owning a car and having a child between the ages of 3 and 7 who rode in the car at least once a week.	Beliefs, behaviours, perceived barriers and motivations.	Participants were recruited from a marketing data base of individuals who met the sampling criteria of owning a car and having a child between the ages of 3 and 7 who rode in the car at least once a week. Institutional Review Board approval was obtained before the study. Written consent was obtained from all participants.	FGDs had a professional group leader with a graduate degree in clinical psychology. The in-depth interviewing of respondents was done with skill in a dynamic atmosphere of rapport, sensitivity and openness to sharing thoughts and feelings. The interview guide contained open-ended questions.	The analysis came from understanding and interpreting the verbal, visual and other inputs from respondents and client-relevant themes. The moderator using the topics identified in advance as the areas of interest completed an in-depth report.
Winston et al (USA)21	To identify factors that influence parents’ current child restraint use behaviours and intentions for future use. To test interventions that address these factors as a means to promote appropriate restraint use behaviours, particularly the use of belt-positioning booster seats.	3 phases: formative phase—FGDs defined barriers; development phase—literature review to identify interventions and in the evaluative phase—4 existing and 3 new interventions were evaluated. There was also a follow-up telephonic survey.	Parents of children between the ages of 3 and 8 years who never or rarely used appropriate restraints. Additional inclusion criteria included: educational attainment of, at most, a high school diploma, and travel with children in motor vehicles at least once a week Formative phase: 10 FGDs, n=117 Evaluative phase: 20 FGDs, n=171	Beliefs and behaviours.	Study participants were recruited in both phases using telephone calls and printed flyers, through an existing injury free children network. All participants were offered reading assistance and/or Spanish language translation, according to the participant’s preference. The Institutional Review Board at The Children’s Hospital of Philadelphia approved all recruitment and consent procedures, as well as all study protocols and surveys.	Discussion guides prepared for both formative and evaluative phases. In evaluative phase, the parent’s reactions to interventions were assessed after they were provided with free belt-positioning booster seats as well as education on their use. Follow-up interviews were conducted 6 weeks later.	FGDs were taped and transcribed. Transcripts were coded, and thematic analyses done by 2 research assistants under the supervision of the study coordinator. When consensus on interpretations was not reached, all interpretations were included in the summary document.
Johnston et al (USA)29	To expand our understanding of the determinants of booster seat use with a broader range of families from our priority population (low-income African American, Somali and Vietnamese parents).	FGDs.	African American, Somali and Vietnamese parents or custodial grandparents of children between birth and 9 years of age and reported that they ‘did not always’ use a booster seat when their booster-eligible child was riding in a car. There were 26 participants across 3 FGDs.	Not specified.	Participants were recruited through posters, flyers and information booths at community clinics, community centres and other social service providers. Interested individuals were prescreened to ensure that eligibility criteria was met. All participants received childcare, a meal, a small stipend ($25) and information on child passenger safety at the conclusion of FGDs.	FGDs were conducted by an experienced African American, male facilitator. Somali and Vietnamese sessions included an interpreter who also cofacilitated the group. A semistructured guide was used. FGDs were conducted at settings familiar and comfortable to the attendees.	FGDs were audiotaped, transcribed and translated. Three researchers independently reviewed transcripts and field notes completed by study staff. Major themes in each language or cultural group were identified and categorised according to the major domains of the theoretical models.
Erkoboni et al (China)22	To explore the efficacy and acceptability of using a US-developed belt-positioning booster seat use promotion intervention in Beijing.	FGDs were conducted in relation to an intervention (a Chinese-produced instructional video and an English-language video dubbed into Mandarin) with theoretically driven messages through a personal story—knowledge only video vs knowledge+motivation video.	Eligible parents were those who owned a car and had children between the ages of 3 and 8 years enrolled in a Beijing kindergarten or elementary school.	Parents’ perceived benefits, disadvantages, facilitators and barriers.	Parents were recruited for the study through kindergartens and elementary schools. Parents were informed about the study in school or parent–teacher meetings and those willing participated (five groups, 71 participants).	Collection of qualitative data on target constructs—initial FGD. Collection of qualitative data on reactions to videos. Parents were shown the two videos and then FGD was conducted after it.	FGDs were taped, transcribed and translated into English. Transcripts were coded by research staff on the basis of the themes. In addition, parent responses to the video interventions were coded to qualitatively assess participant’s response to the programmes. To supplement these transcripts, field notes were taken by researchers present during the discussion. A translator was present during the FGD for each English-speaking researcher.
Brown et al (Australia)30	To qualitatively explore barriers to optimal child restraint use using the integrative behaviour change model in culturally and linguistically diverse communities in New South Wales, Australia.	11 language specific FGDs involving 71 parents or grandparents of children aged 3–8 years.	Participants had to (1) speak a language other than English at home; (2) be aged over 18 years and (3) have travelled in a car with at least one child between 3 and 8 years old in the last 6 months.	Existing knowledge and experience as well as underlying beliefs and motivations.	71 participants for 11 FGDs. Recruitment was done using advertisements in target languages through existing community groups or by direct invitation from community leaders to increase homogeneity within groups, and ensure participants were in comfortable and familiar settings to encourage free discussion.	FGDs were held over a 2-hour period and aided by a semistructured topic guide which was not changed during the entire study. A standardised questionnaire was used to collect demographic information.	Content analysis using major concepts of Fishbein’s integrative behaviour change model.42 Initially, a single researcher went through the transcripts to do content analysis. This was repeated by a second author and finally two other researchers reviewed the final assignment of discussion elements to the model constructs. Non-spoken communication was not communicated in the study.
Chen et al (China)33	To explore parent drivers’ perceptions and experiences regarding use of child safety restraint, and to identify the differences in the perceptions and experiences between child safety restraint users and non-users.	14 IDIs among parents with a child under the age of 6 years. Of 14 parents interviewed, 7 parents were child safety seat users while 7 were non-users. Ages of parents ranged from 29 to 34 years.	Parents living in downtown Shantou, with a child under the age of 6 years living in the same household, and who had been a primary driver and driving more than one time every week for at least 2 years.	Perceptions and experiences regarding use of child safety restraint.	20 participants who met the selection from the 1069 drivers who had previously participated in an observational survey, and expressed willingness to be contacted for future study were identified for IDIs. A small gift of child stationery and a toy were given.	Face-to-face interviews were conducted by one trained author. Each of the interviews lasted 35–50 min and was guided by a topic guide designed based on the Health Belief Model. All of questions were open-ended.	Research team familiarised itself with the data and to identify patterns and their links to the theoretical framework. Thematic analyses were done using NVivo V.10. Finally, the bilingual and bicultural research team evaluated the data and results.
Nelson et al (Saudi Arabia)31	To identify the salient beliefs of pregnant women in Saudi Arabia that may influence their intentions to use infant restraints for their children.	FGDs.	25 pregnant women participated in 2 FGDs.	Behavioural, normative and control beliefs were elicited.43	Pregnant women from Dallah Hospital were eligible to participate regardless of the number of children. The hospital’s representatives approached individuals during their regular prenatal appointments with an invitation to participate in the study and explained the requirements. Informed consent was obtained prior to the start of each of the FGDs. participants received a gift card equivalent to $13.33.	The FGDs were moderated by the health educator from the research team and were digitally recorded, after which they were transcribed and translated into English by an independent translator. A topic guide was used.	Researchers used the cut-and-paste technique, where the transcript of the discussions was cut up into segments and grouped into individual constructs. Statements and comments of each individual section (behavioural, normative and control belief) were then analysed and sorted into specific logical categories. Each category was then analysed and a belief statement was formulated.
Liu et al (China)35	To investigate the knowledge, attitudes and intended behaviours of child safety seat use among parents of newborns and to explore expectant mothers’ views and decisions regarding child safety seat use.	Mixed-method cross-sectional survey and semistructured interview.	The inclusion criteria were: >12 weeks pregnant; owned a car and agreed to participate through signed consent. Those with high risk pregnancy or who were unable to complete more than 15 min of an in-person interview were excluded. The targeted sample size was 30 to reach saturation.	Views and decisions regarding child safety seat use.	Pregnant women who sought prenatal care at Shantou Women’s and Children’s Hospital were recruited for a semistructured interview. Nothing else reported. The study protocol, along with the consent process, was approved by Medical Ethics Committees of Shantou University Medical College.	A semistructured interview guide was used. Interviews were conducted in a private room in the hospital and lasted about 30 min. All interviews were audio recorded and transcribed into electronic documents word-for-word on the same day by another author of this manuscript before data analysis.	The transcribed electronic documents were imported into NVivo V.8.0 and read repeatedly by two researchers, who coded and analysed them thematically. The data and results were assessed by researchers who speak English, Chaozhou-Shantou dialectal and Mandarin, the languages used by the participants.
Fleisher et al (USA)23	To assess perspectives on and use of existing tools to guide car seat installation, as well as to explore the acceptability of using a mobile app to guide car seat installation.	Mixed-methods study consisting of survey, and FGDs using phenomenological approach.	Parent/guardian with a child between ages 0 and 5 years (inclusive) treated at the participating practice. Participant eligibility was determined according to the age of the child, child race, child medical insurance type (private or Medicaid), parent report of transporting a child in a car seat at least 2 times per week, parent report of installing any type of car seat at least 5 times in the last 6 months, and parent ability to travel to the local paediatric practice.	Perspectives and experiences.	Primary care practices were chosen based on geographic location to ensure variability in participant race/ethnicity, community setting and socioeconomic status. A list of 400 potentially eligible participants was generated for each practice. Eligible parents/caregivers were invited to participate by mail and telephone. Parents and caregivers who contacted or were contacted by study staff and screened to be eligible were enrolled and scheduled for a focus group until an initial enrolment goal of at least 40 participants was achieved.	The lead author who has training and extensive experience in focus group facilitation led the FGDs. The participants were recruited through each primary care practice and were not familiar with the research team. The FGDs were conducted at each of the participating clinics and therefore convenient for participants. A phenomenological research approach was employed.	FGD recordings were transcribed and analysed using inductive thematic analysis to identify patterns across the data set. Three reviewers conducted initial coding and then coding conferences were held to discuss and resolve discrepancies. In the first phase of analysis, apparent themes were identified from an initial review of the data. FGD recordings were then transcribed and read in entirety by members of the research team, who identified initial codes. The transcripts were then imported to NVivo qualitative software where data was further coded and themes and subthemes across the dataset were identified and refined. Data were once again read in their entirety, along with accompanied memos, to ensure the themes identified were congruent with dataset and portrayed participants’ experience of the phenomenon. Finally, themes were defined, categorised and named.
Hunter et al (Australia)36	To explore parental knowledge and attitudes and self-reported child car seat use.	Mixed-methods study consisting of survey, and FGDs using the PRECEDE-PROCEED conceptual framework.	Parents or carers whose children aged 3–5 years were enrolled in any of the 3 early learning centres (two preschools and one long day care centre) in regional NSW. 10 in 3 FGDs.	Knowledge and attitudes.	Participants were recruited through posters that were displayed at the centres. The centre director was also asked to encourage parents of Aboriginal children to attend and directly approached participants through personal invitation.	A researcher experienced in qualitative methods conducted three FGDs of 60–90 min duration at each of the centres, 6–14 weeks after the surveys were administered. Focus groups were conducted during the day at times convenient for parents and centre staff.	FGD recordings were listened to in full then transcribed verbatim by a transcription service and then checked for accuracy by the first author. Each transcription was read and reread and codes were developed deductively, based on the theoretical framework. Non-verbal communication was not systematically interpreted. NVIVO V.9 to organise the qualitative data.
Mckenzie et al (USA)37	To determine prevalence of child safety seat use, awareness of and barriers to use; and to use the information gained in the first phase to develop, implement and evaluate a child passenger safety programme for Somali families living in Columbus, Ohio that is both sustainable and culturally appropriate.	Mixed-methods study consisting of survey, IDIs and 2 FGDs to develop an intervention followed by a telephonic survey to get feedback on the intervention. The intervention was a video which was given to community leaders to show—this was developed based on results of a component of this study.	Key informants (10) in the local Somali community were initially identified by the executive director of a Somali family alliance organisation in Columbus, Ohio. Additional community leaders were identified as the interviews progressed. To participate in FGDs, parents had to be Somali, at least 18 years of age, and have at least one child ≤8 years old.	Knowledge of and barriers to proper use in order to inform development, implementation and initial evaluation of a culturally appropriate intervention for Somali families.	Key informant (n=10) were initially identified by the executive director of a Somali family alliance organisation and later snowballing done. No information is provided on how Somali parents (n=30) were recruited. The study was approved by the Institutional Review Board at the Research Institute at Nationwide Children’s Hospital.	No information provided on data collection of key informants, but for FGD, it is mentioned that an interpreter was used, and open-ended questions were asked.	Only mentions that the data were recorded and transcribed and common themes were extracted on review and briefly summarised.
Hall et al (Australia)32	To explore how child restraint use might function as part of an individual’s capability, opportunity and motivation and how this varies across high education and income (high SES), low education and income (low SES) and culturally and linguistically diverse (CALD) users.	FGDs.	All parents or carers over 18 years and conversant in English were eligible. Two groups each of high SES participants, CALD participants and low SES participants. All participants had some previous experience in installing a child restraint system or securing a child into a restraint. Participants were reimbursed AUD$25.00 for their travel costs.	Perceptions (insights) and experiences.	High SES participants were recruited through university and research organisation email distribution channels. Community playgroups specifically for English as second language residents in South-eastern Sydney, and community playgroups in areas low SES as indicated by the Australian Government were used to recruit CALD and low SES participants.	Two researchers attended each FGD; one facilitated discussion using a semistructured discussion guide and the other took notes. The discussion guide was formulated on a review of factors previously reported to be associated with errors in child restraint use.	The COM-B model of behaviour was used to deductively analyse discussions. Transcripts and discussion notes were then coded according to these categories independently by two researchers using QSR International's NVivo V.11 Software. This included any mention of any issue relevant to these predefined categories. Where any consistencies between researchers occurred, these were discussed between the researchers until consensus was reached. Numerical counting was done of codes.